Open Collections

Tracking and recognizing actions of multiple hockey players using the boosted particle filter

University of British Columbia

This thesis presents a system that can automatically track multiple hockey players and simultaneously recognize their actions given a single broadcast video sequence, where detection is complicated by a panning, tilting, and zooming camera. Firstly, we use the Histograms of Oriented Gradients (HOG) to represent the players, and introduce a probabilistic framework to model the appearance of the players by a mixture of local subspaces. We also employ an efficient offline learning algorithm to learn the templates from training data, and an efficient online filtering algorithm to update the templates used by the tracker. Secondly, we recognize the players' actions by incorporating the HOG descriptors with a pure multi-class sparse classifier with a robust motion similarity measure. Lastly, we augment the Boosted Particle Filter (BPF) with new observation model and template updater that improves the robustness of the tracking system. Experiments on long sequences show promising quantitative and qualitative results, and the system can run smoothly in near realtime.

Text

2011-02-28

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/31853

Computer Science

University of British Columbia

Graduate